EUVD-2026-82973
GHSA-4mwh-5hjw-p5jv
Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessCraft HTML artifact with malicious JS
-
DeliveryPublish via public Share view
-
ExploitSend share link to victim
-
ExecutionVictim opens link, Preview renders payload
-
ImpactScript executes in Kimi origin, hijacks session
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A05:2025 - Injection
ASVS 5.0
V1 - Encoding and Sanitization Output encoding, context-aware escaping
MITRE ATT&CK
2 techniques across 2 tactics: Initial Access, Execution The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Exploitation requires the attacker to create a crafted HTML artifact and publish it through Kimi's public Share view; the malicious payload triggers when the artifact's Preview rendering component fails to sanitize/escape HTML, executing attacker JavaScript in the Kimi application origin. The key limiting factor is required user interaction (UI:R): a victim must open the attacker-supplied share link - there is no zero-click path. No authentication is needed to publish the share (PR:N), but the affected feature (public HTML artifact Preview / Share view) must be in use, and impact is confined to the victim's browser session in the Kimi origin (S:C, C:L/I:L). |
| Impact | This is a client-side stored/reflected Cross-Site Scripting flaw (CWE-79) in the Kimi web application (Moonshot AI), specifically in how the HTML artifact Preview renders content within the public Share view. The provided CVSS 3.1 vector AV:N/AC:L/PR:N/UI:R/S:C/C:L/I:L/A:N yields 6.1 (Medium), which is consistent with a classic XSS: network-reachable, low complexity, no privileges, but requiring victim interaction (UI:R - opening a malicious shared link) and a scope change (S:C - attacker script runs in the Kimi application origin rather than the attacker's). The 'execute arbitrary code' phrasing in the description means arbitrary JavaScript in the victim's browser session, NOT server-side RCE - the confidentiality/integrity impact is Low (session-bound), and availability is unaffected (A:N), so the 'RCE' tag overstates the true impact. A GitHub reference (MGTx2/CVE-2026-79294) points to a PoC repository, so publicly available exploit code exists. No EPSS score, no CISA KEV listing, and no SSVC decision data were provided in the input. Net: a real but moderate-priority issue - no evidence of active exploitation, and exploitation depends on social-engineering a victim into viewing a crafted share link; treat it as a standard web-app XSS to remediate, not an emergency. |
| Exploit ScenarioDerived | An attacker crafts a Kimi HTML artifact containing malicious JavaScript and publishes it via the public Share feature, then distributes the share URL. When a victim opens the shared link, the Preview rendering executes the attacker's script in the Kimi origin, allowing theft of the victim's session tokens or actions on their behalf. A PoC is referenced on GitHub (MGTx2/CVE-2026-79294), and the low attack complexity (AC:L) plus network vector (AV:N) make delivery trivial, though it requires the victim to open the link (UI:R). |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Sep 18, 2026
Cross-site scripting in Moonshot AI's Kimi web application (kimi.com, affected as of 2026-07-18) lets a remote, unauthenticated attacker publish a crafted HTML artifact through the public Share view so that the artifact Preview rendering path executes attacker-supplied JavaScript in the Kimi origin. The flaw is exploitable only against a victim who opens the attacker's share link (CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:C/C:L/I:L/A:N, CVSS 6.1) - there is no zero-click path, and impact is confined to the victim's browser session in the Kimi origin (S:C with limited confidentiality and integrity effects). The references include a public GitHub repository (MGTx2/CVE-2026-79294) that appears to document the issue, so proof-of-concept material may be publicly accessible; there is no CISA KEV entry and no confirmation of active exploitation at time of analysis. Note that although the CVE description says 'execute arbitrary code,' the assessed impact is arbitrary client-side JavaScript execution, not server-side code execution.
RemediationAI
No vendor-released patch identified at time of analysis, so rely on compensating controls. Because this is a hosted service, Kimi/Moonshot AI must fix it on the server side: sanitize and escape HTML artifact content with a vetted sanitizer (e.g., DOMPurify) before rendering in Preview, and render shareable artifacts inside an isolated origin - a sandboxed iframe with an explicit sandbox attribute omitting allow-same-origin, or a separate artifact-only domain - so that any residual script cannot execute in the kimi.com origin; pair this with a strict Content-Security-Policy (nonce/hash-based script-src, no unsafe-inline) and a Trusted-Types policy on the Share view. The trade-off is that aggressive sanitization or sandboxing will break legitimate interactive HTML artifacts and embedded scripts in shared previews, so artifact authors should be told which features are unsupported. Until a fix is confirmed, end users and administrators should avoid opening Kimi share links from untrusted parties, and organizations should restrict or disable sharing of HTML artifacts via the public Share view if policy allows; note that blocking the /share/ path reduces collaboration functionality, and the primary residual risk is user interaction, since a victim must actively open the crafted link. Monitor the vendor sites (http://kimi.com, http://moonshot.com) and the reference at https://github.com/MGTx2/CVE-2026-79294/blob/main/README.md for updated guidance; no patched version number can be cited because none is provided in the available data.
Affected VersionsENISA
Affects your stack
Find out whether the stacks and servers you run are on an affected version.
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
The root cause is CWE-79 (Improper Neutralization of Input During Web Page Generation), realized as a client-side stored/reflected XSS in Kimi's HTML artifact Preview component inside the public Share view. Kimi lets users generate and publish HTML artifacts that render in a preview pane; when a shared artifact is viewed, the application fails to sanitize or escape embedded markup, so attacker-controlled script is emitted into the Kimi application origin rather than into a hardened, sandboxed rendering context. Because the payload lives inside a shareable artifact that persists for subsequent viewers, this behaves like stored XSS delivered via a link, with the scope change (S:C) reflecting that the script executes with the privileges of the Kimi origin rather than the artifact's own contained context. The supplied CPE string (cpe:2.3:a:n/a:n/a) and the EUVD affected-version list ('n/a n/a') carry no product or version granularity, so the exposure must be scoped by the service endpoints in the references - kimi.com and the share route at kimi.com/share/ - rather than by a versioned software package. The CVSS vector confirms a network-reachable, low-complexity, unauthenticated attack that still requires victim interaction (UI:R) and yields partial confidentiality/integrity impact with no availability impact.
Know when the next one hits your stack
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
AV:N/AC:L/PR:N/UI:R/S:C/C:L/I:L/A:N
Standard reflected/stored XSS: network vector, no privileges to post a share, victim must open the link (UI:R), script escapes into the app origin (S:C), limited C/I impact, no availability effect.
Primary rating from Vendor (mitre).
CVSS VectorVendor: mitre
DescriptionCVE.org
Cross Site Scripting vulnerability in Moonshot AI Kimi version as of 2026-07-18 allows a remote attacker to execute arbitrary code via the HTML artifact Preview rendering; public Share view component
Threat Intelligence Overview
NVD
CVE.org
EPSS
Vulnrichment
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
Cross-Site Scripting occurs when an application accepts untrusted data and sends it to a web browser without proper validation or encoding.
Exploit-DB